Investigation of working memory during semantic and syntactical processes with functional magnetic resonance imaging

There is limited neuroimaging evidence on distinguishing between semantic and syntactic domains in terms of verbal working memory (vWM). The current study investigated brain activations and functional connectivity patterns during semantic and syntactic vWM via functional magnetic resonance imaging (fMRI). For this purpose, semantic vWM was explored with effects of word frequency and load, whereas syntactic vWM with effects of load, and two syntactic elements, binding and movement. Data from 25 healthy participants revealed activation mainly in dorsal attention (DAN), control (CONT) and visual (VIS) networks. Default-mode (DMN) and salience/ventral attention networks deactivated during both tasks. Prominent activations were observed in bilateral inferior frontal gyri (IFG) during semantic, whereas in left fronto-parietal regions, cerebellum and thalamus during syntactic vWM. Word frequency produced increased activation in bilateral DAN, CONT and VIS in semantic vWM. Left IFG activation increased during binding condition compared to movement in syntactic vWM. Right IFG activation increased in overall semantic condition compared to syntax. Network-based statistics on functional connectivity revealed a configuration mostly among DMN, DAN, CONT and VIS nodes. These network nodes exhibited increased connectivity in semantic condition compared to syntax, whereas in syntax, compared to semantic, DAN and DMN showed increased within network connectivity across hemispheres. Also, left temporo-parietal regions were strongly involved. Our results indicate that bilateral IFG are involved in semantic vWM and showed increased activation related to cognitive load. Left fronto-temporo-parietal regions are involved in syntactic vWM and several right hemisphere nodes of networks exhibited connectivity alterations regardless of changes in activation. Key Words: Functional magnetic resonance imaging, working memory, linguistics, neuroimaging, short-term memory The present work was supported by TÜBİTAK. Project No. 114E581